A narrated, hands-on workshop · ~60 min · for people who will never write code

AI Without Code

Press play and I'll show you what's actually inside the machine — how it reads, how it remembers, why it rolls dice, and why it sounds so sure when it's wrong. Zero programming. Just the ideas that make you dangerous with it.

13 chapters · player controls live at the bottom

TOKEN LAB — what the machine actually sees approximate

Type anything, or pick a preset above.

Honest label: this splitter is a teaching sketch — every real AI tool slices text with its own trained tokenizer, so real counts differ. The idea it shows (words become sub-word pieces; long words, numbers and non-English text cost more pieces) is the real thing.

Ch. 1 / What it actually is ▶ narrated

It's not a library. It's a prediction machine.

Forget the robot from the movies and forget the all-knowing database. The machine behind every AI chat window does exactly one thing, over and over: given everything so far, it predicts a likely next little piece of text. That single fact explains nearly everything you'll ever see it do — the brilliance and the failures.

📚

Not a lookup

It doesn't fetch answers from a shelf of stored documents. There's no file for your question. It generates every reply fresh, piece by piece.

// no shelf, no file

🎲

A next-piece guesser

Trained on oceans of text, it learned which pieces tend to follow which. Your reply is a chain of likely next pieces — chosen live, in front of you.

// prediction, not recall

✨

Why that's still amazing

Predicting text well turns out to require absorbing patterns of grammar, style, reasoning and world knowledge. The skill is real. The certainty is not.

// skill ≠ certainty

Hold this line for the whole hour: fluent is what it always is; correct is what it sometimes is. The rest of the course teaches you to tell the two apart.

Ch. 2 / How to treat it ▶ narrated

Meet your intern: brilliant, forgetful, shameless.

The healthiest way to hold this tool is as a very specific kind of colleague. Three quirks define the whole relationship — two you manage around, one you exploit.

🫥

No memory

Every new chat starts from zero. It knows nothing about you, your job, or yesterday's conversation unless you put it in the message.

// you carry the context

😶

No shame

It never says "I'm not sure" on its own. A wrong answer arrives in the same confident, polished voice as a right one. Always.

// confidence is a costume

🔋

No fatigue

It gives the thousandth request the same effort as the first, at any hour, without sulking. This is the quirk you exploit ruthlessly.

// stamina is the superpower

The division of labor that falls out: it drafts, transforms and organizes — you judge, verify and decide. Keep those lanes and you can use it aggressively without ever getting burned.

Ch. 3 / The first skill ▶ narrated

The good ask: four parts, no code.

The single fastest upgrade to every answer you'll ever get is not a secret phrase — it's structure. Context, task, format, constraints. Click each scenario below and watch a lazy ask become a professional one.

The lazy ask

The good ask

      context — who/what this is
      task — the exact job
      format — the shape you want back
      constraints — the guardrails
    

Do this now: click through all three scenarios and read the strong version out loud. Notice you could write every one of them today — there's nothing technical in any of them. That's the point.

Ch. 4–12 / Locked

That's the free preview.

You've heard the setup. The remaining 9 chapters are where AI Without Code actually gets built — the guardrails, the hands-on builds, the full narrated walkthrough and the end-of-workshop check.

Get the full workshop → See the catalog

One file. Yours forever. Works offline, in any browser, on any device.